SQL 从一列返回值的查询是重复的，但在另一列中有 1 null 和一个值

Question

Important: Metabase SQL is READ only, this metabase is using a POSTGRESQL.

The query I need is: Return all the values in Shipment that in the shipment col has the same number and in the col Origin has 1 null and 1 value (coupled).

This is my table:

Origin	Shipment	Product	Quantity
Purchase.23	shipment.42	356	2
	shipment.42	356	2
Purchase.25	shipment.422	256	4
	shipment.422	256	4
Purchase.18	shipment.427	666	4
	shipment.427	666	4
Purchase.20	shipment.427	666	3
	shipment.427	666	3
Purchase.22	shipment.427	677	4
	shipment.427	677	3
	shipment.9999	222	9
	shipment.427	677	1
	shipment.9999	222	9
	shipment.1111	7	11
	shipment.1111	564	6
	shipment.5555	756	1
	shipment.4444	755	5
Purchase.77	shipment.3333	751	9

Notice in the table above, that Purchase.22 has 4 in Quantity and there are two shipments with the same shipment as the Purchase.22 row with 1 and 3 in quantity, both summed are the 4 of the Purchase.22 row.

The query should return this:

Origin	Shipment	Product	Quantity
Purchase.23	shipment.42	356	2
	shipment.42	356	2
Purchase.25	shipment.422	256	4
	shipment.422	256	4
Purchase.18	shipment.427	666	4
	shipment.427	666	4
Purchase.20	shipment.427	666	3
	shipment.427	666	3
Purchase.22	shipment.427	677	4
	shipment.427	677	4

And if posible another query that returns this, products that doesn't have any couple Origin null with Shipment.xx, Origin Purchase with Shipment.xx:

Origin	Shipment	Product	Quantity
	shipment.9999	222	9
	shipment.9999	222	9
	shipment.1111	7	11
	shipment.1111	564	6
	shipment.5555	756	1
	shipment.4444	755	5
Purchase.77	shipment.3333	751	9

A little bit more of info, I'm working on a BI tool named Metabase, I'm a junior I just started a month ago and they gave me this stuff to do and the query I'm making is to know how many articles I have on the ivnentory. Unfortunately some articles are duplicated, if they were all duplciated I could easily divide the quantity by 2. But i's never easy. I have values that completly lack of origin and values that lack of shipment and both aren't coupled with other value that has both, and other products that are triplicated and have different quantitites, hence I cannot divide it that easily.

I tried to find any pattern with other columns form the table, but nothing worked at least on shipment.out while in other shipment values like shipment.out.return I found patterns that helped me without the necessity of using a SQL query.

I tried to find any pattern with other columns form the table, but nothing worked at least on shipment.out while in other shipment values like shipment.out.return I found patterns that helped me without the necessity of using a SQL query.

Answer 1

First you need to aggregate your quantities across <shipment, product>.

Then you can select:

• the last value for each <shipment, product, quantity>
• the count of each <shipment, product, quantity>

The filter out all rows that have less than 1 count and which have both <origin, last origin> set to null.

Check the demo here .

---

If you can have multiple records for <shipment, product> and need to aggregate quantity, you need to do it before the execution of this query. Here's how it would look like:

WITH cte AS (
    SELECT MAX(Origin) AS Origin, Shipment, Product, SUM(Quantity) AS Quantity
    FROM tab
    GROUP BY COALESCE(Origin, 'NULL'), Shipment, Product
), cte2 AS (
    SELECT *, 
           LAG(Origin) OVER(PARTITION BY Shipment, Product, Quantity ORDER BY Shipment, Origin) AS last_val, 
           COUNT(*) OVER(PARTITION BY Shipment, Product, Quantity) AS cnt
    FROM cte
)
SELECT Origin,
       Shipment,
       Product, 
       Quantity
FROM cte2
WHERE COALESCE(Origin, last_val) IS NOT NULL AND cnt > 1
ORDER BY shipment

Output:

origin	shipment	product	quantity
Purchase.23	shipment.42	356	2
null	shipment.42	356	2
Purchase.25	shipment.422	256	4
null	shipment.422	256	4
Purchase.20	shipment.427	666	3
null	shipment.427	666	3
Purchase.18	shipment.427	666	4
null	shipment.427	666	4
Purchase.22	shipment.427	677	4
null	shipment.427	677	4

Check the demo here .